Skip to content

More than A Trillion Ways to Slice Your Data

September 7, 2016

At Evans Data we do literally dozens of in-depth developer surveys every year.  There’s ten focused surveys of our own, plus typically twenty to thirty custom research projects for private clients.  Most surveys are fairly long and technical and all have at least 400 respondents (that’s where you get the industry standard of plus or minus 5% margin of error).  Many of the surveys are worldwide with regional or country quotas that then build up to thousands of developers.  That’s a lot of developer data!


And with all that data we inevitably find that clients want it cut in various ways that we wouldn’t have thought of.  They might want to look at something so broad as all the results cut by company size or platform usage.  Or it might become so specific that they want to look at and profile only those developers who reside in Brazil, work for companies with more than 1000 employees, target Android, and use Java.  Or maybe they want to see the crossover between mobile and IoT development, or maybe they want know which region and industry is the leading area for machine learning implementations.


As you might guess, the quest for just the right type of data can get dicey.  As data gets sliced thinner and thinner the margin of error increases, although with enough data this doesn’t become a problem until it gets extreme.  We do provide guidance along those lines


However, the other thing that a relentless desire for unique views into the data brings is a large time sink.  Manipulating the data manually takes many man-hours which is costly in terms of time, resources, and money.  And manual labor makes little sense for a technology company, so we’ve come up with a Data Analytics Console that enables the end-user clients to drill down on any item in any configuration they want.  Currently available for our Global Development survey series, it takes over 160 raw data points (questions) provides for 4 different crosstab for each one and allows for 26 different filters to be applied.  Multiple filters can be selected and multiple combinations of filters can be implemented at the same time. To count the number of combinations, we have to compute the number of permutations 26 filters can be arrayed, while accounting for the fact that a user can choose one filter, two filters, three filters, anywhere up to 26 filters at the same time, This allows for 42,949,672,320 total possible charts – though applying too many filters will give you unusable results, which is why we urge caution and restraint.


Check it out here:

No comments yet

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: